Discussion:
gdbstub initial code, v15
Oleg Nesterov
2010-11-01 17:59:27 UTC
Permalink
I just realized I forgot to send V15 before (I was distracted by
sigaltack bugreport).

Changes: multiple watchpoints.
Any advice is very much appreciated. Most probably, there is no any
clever solution. Once a traced sub-thread detects that a watchpoint
was changed, it should mark this wp as "reported" for other threads
and report it to gdb. IOW, we report the random thread and random wp.
This is what ugdb does. I don't think we can do something better
without changing the remote protocol.

Unfinished: support "signal SIG" in T00 case correctly (without
another report to gdb).

Next: hardware watchpoints. But this is tricky, and right now I am
not sure ugdb can do this (at least unless we ignore conflicts with
other users of debugregX/ptrace_bps).

Oleg.
Kevin Buettner
2010-11-01 18:23:48 UTC
Permalink
On Mon, 1 Nov 2010 18:59:27 +0100
Post by Oleg Nesterov
Next: hardware watchpoints. But this is tricky, and right now I am
not sure ugdb can do this (at least unless we ignore conflicts with
other users of debugregX/ptrace_bps).
For the time being, I would just ignore those conflicts. Later on,
perhaps, it might make sense to implement facilities by which the
various users can lock the resources in question.

Kevin

Loading...